The proposed project will test for association between three phenotypes related to alcohol dependence and abuse: lifetime alcohol dependence, novelty seeking and childhood externalizing behavior; and seven candidate genes that have been shown in previous studies to be implicated in alcoholism and related phenotypes. These studies will utilize a sample of approximately 1300 individuals who have been studied longitudinally since their initial enrollment in the National Youth Survey (NYS) in 1976. The sample is a nationwide probability sample of the United States in that year. Since this sample is from a widely diverse population, population stratification and the possibility of spurious associations are legitimate concerns. Therefore, analyses must incorporate population stratification information in case-control association tests. We will genotype 100 unlinked short tandem repeat (STR) markers spanning the autosomal genome and use these 100 markers to detect population stratification by applying recently developed population genetic methods. Assuming population stratification is present in this diverse sample, analyses will be performed to characterize and incorporate population stratification information into case-control association tests of each of the three phenotypes against each of the seven candidate genes. Because the sample size is large, it will be divided into two groups, providing an opportunity for replication. Furthermore, results of these case-control association tests can be compared to family-based tests that will become possible in a subsequent multigenerational phase of the National Youth Survey. Thus, the results from two different methods applied to the same sample can be compared. This proposed study will demonstrate the application of a powerful and efficient case-control approach to the study of genetic association, taking proper account of population structure, in a representative sample of the United States. This will open the way to large-scale genetic studies of significant public health problems, such as alcoholism, without the need for expensive and often infeasible pedigree-based research designs.